Rapid detection of phase transitions from Monte Carlo samples before equilibrium

TL;DR

This paper introduces a deep learning approach using Bidirectional LSTM and Transformer models to detect phase transitions from Monte Carlo data before equilibrium, significantly reducing computational costs.

Contribution

It demonstrates that deep learning models can classify phases and identify transition points using raw data prior to equilibrium, a novel application in condensed matter physics.

Findings

Deep learning models accurately classify phases from raw Monte Carlo data.

The method detects phase transitions, including Kosterlitz-Thouless, before equilibrium.

Performance is robust across different data types and models.

Abstract

We found that Bidirectional LSTM and Transformer can classify different phases of condensed matter models and determine the phase transition points by learning features in the Monte Carlo raw data before equilibrium. Our method can significantly reduce the time and computational resources required for probing phase transitions as compared to the conventional Monte Carlo simulation. We also provide evidence that the method is robust and the performance of the deep learning model is insensitive to the type of input data (we tested spin configurations of classical models and green functions of a quantum model), and it also performs well in detecting Kosterlitz-Thouless phase transitions.